mirror of
https://github.com/james-m-jordan/morphik-core.git
synced 2025-05-09 19:32:38 +00:00
5d76521059
* add filename option for text documents * fix: duplicated env variables * Include chunk score in query response * don't initialize colpali store if one is not provided * Add score to ChunkSource * Add huggingface cache volume to avoid redownloads * Replace <~> in maximum_similarity calculation * Add dev_mode setup to default configuration * Dump with databridge documents (no pdfs) * Add dev_mode setup to default configuration * Comment docker instructions for api host * Use empty dump file
111 lines
2.4 KiB
TOML
111 lines
2.4 KiB
TOML
[api]
|
|
host = "localhost" # Needs to be "0.0.0.0" for docker
|
|
port = 8000
|
|
reload = true
|
|
|
|
[auth]
|
|
jwt_algorithm = "HS256"
|
|
dev_mode = true # Enabled by default for easier local development
|
|
dev_entity_id = "dev_user" # Default dev user ID
|
|
dev_entity_type = "developer" # Default dev entity type
|
|
dev_permissions = ["read", "write", "admin"] # Default dev permissions
|
|
|
|
[completion]
|
|
provider = "ollama"
|
|
model_name = "llama3.2-vision"
|
|
default_max_tokens = "1000"
|
|
default_temperature = 0.7
|
|
# base_url = "http://ollama:11434" # Just use the service name
|
|
base_url = "http://localhost:11434"
|
|
|
|
# [completion]
|
|
# provider = "openai"
|
|
# model_name = "gpt-4o"
|
|
# default_max_tokens = "1000"
|
|
# default_temperature = 0.7
|
|
|
|
[database]
|
|
provider = "postgres"
|
|
|
|
# [database]
|
|
# provider = "mongodb"
|
|
# database_name = "databridge"
|
|
# collection_name = "documents"
|
|
|
|
[embedding]
|
|
provider = "ollama"
|
|
model_name = "nomic-embed-text"
|
|
dimensions = 768
|
|
similarity_metric = "cosine"
|
|
# base_url = "http://ollama:11434" # Just use the service name
|
|
base_url = "http://localhost:11434" # Local configuration
|
|
|
|
# [embedding]
|
|
# provider = "openai"
|
|
# model_name = "text-embedding-3-small"
|
|
# dimensions = 1536
|
|
# similarity_metric = "dotProduct"
|
|
|
|
[parser]
|
|
chunk_size = 1000
|
|
chunk_overlap = 200
|
|
use_unstructured_api = false
|
|
use_contextual_chunking = false
|
|
|
|
[parser.vision]
|
|
provider = "ollama"
|
|
model_name = "llama3.2-vision"
|
|
frame_sample_rate = -1 # Set to -1 to disable frame captioning
|
|
base_url = "http://localhost:11434" # Only used for ollama
|
|
# base_url = "http://ollama:11434" # Use if using via docker
|
|
|
|
# [parser.vision]
|
|
# provider = "openai"
|
|
# model_name = "gpt-4o-mini"
|
|
# frame_sample_rate = -1 # Set to -1 to disable frame captioning
|
|
|
|
[reranker]
|
|
use_reranker = true
|
|
provider = "flag"
|
|
model_name = "BAAI/bge-reranker-large"
|
|
query_max_length = 256
|
|
passage_max_length = 512
|
|
use_fp16 = true
|
|
device = "mps"
|
|
|
|
# [reranker]
|
|
# use_reranker = false
|
|
|
|
[storage]
|
|
provider = "local"
|
|
storage_path = "./storage"
|
|
|
|
# [storage]
|
|
# provider = "aws-s3"
|
|
# region = "us-east-2"
|
|
# bucket_name = "databridge-s3-storage"
|
|
|
|
[vector_store]
|
|
provider = "pgvector"
|
|
|
|
# [vector_store]
|
|
# provider = "mongodb"
|
|
# database_name = "databridge"
|
|
# collection_name = "document_chunks"
|
|
|
|
[rules]
|
|
provider = "ollama"
|
|
model_name = "llama3.2"
|
|
batch_size = 4096
|
|
|
|
[databridge]
|
|
enable_colpali = true
|
|
|
|
[graph]
|
|
provider = "ollama"
|
|
model_name = "llama3.2"
|
|
|
|
# [graph]
|
|
# provider = "openai"
|
|
# model_name = "gpt-4o-mini"
|